The State of Costing: what 33 organizations told us about how they really cost.

One dimension stands up. The rest are bunched.

Average score by dimension, out of 100 (n = 33 for every dimension). The dashed line marks the 50 midpoint. Source: 33 external Profit Check self-assessments, 5 to 15 May 2026.

A small sample, mostly small companies, mostly Portugal

Of the 24 organizations that stated a size, 20 have fewer than 51 people; 9 did not state one. 29 of the 33 took the Portuguese edition of the Profit Check and 4 took the English one, so this is in practice a reading of one market and not of Europe. 17 gave a business email address, 3 gave a personal one, and 13 completed the assessment without giving an address at all; all 33 named their organization, which is how we count one response per organization.

Three industries reach the five-organization threshold we set for publishing a cell: retail (6 organizations, averaging 37), professional services (6 organizations, averaging 29.3), healthcare (5 organizations, averaging 41.6). Four further industries had fewer than five responses and 5 organizations chose "other", so those 16 are counted in every total on this page and are not reported as a cell. We publish no cut of industry by size, and no cut of industry by dimension, at this sample size.

1. Across 33 organizations assessed between 5 and 15 May 2026, the median score is 33 out of 100 and the mean is 34.1. Three organizations scored above 50.
2. Profitability visibility (46.9) is the only dimension clearly ahead of the others.
3. The four lowest dimensions sit between 28.3 and 29.4. At this sample size we do not claim an order among them.
4. 30 of 33 organizations (91%) score 50 or below. The middle half of the sample falls between 22.5 and 45.
5. The lowest overall score was 12 and the highest was 67.
6. The shape holds: organizations can see a margin, and have not built the model that would make the margin defensible.

A wider base, a rule we can state, and one claim withdrawn

1. The first edition of this report, published on 10 July 2026, counted 19 organizations. It counted only those who gave a business email address. This edition counts every completed assessment from a named organization outside our own team, whether or not an address was given, which is why n is 33.
2. Two responses in the first edition do not survive the rule stated below. One carried a test marker in place of a company name. One was taken after the assessment was revised, and is no longer comparable.
3. The Profit Check was revised at the start of June 2026 and one of its seven dimensions changed. Responses collected from 4 June onward are therefore measured on a different instrument and are not pooled here. There are five of them, which is below our publishing threshold, so the next edition of this report waits on the assessment itself rather than on us.
4. The first edition said process and capacity design was the weakest dimension at 25.5. On the wider base it is 28.3, and three other dimensions are within about one point of it. That claim is withdrawn and replaced with the group reading above.
5. The collection window is narrower than the first edition stated, because the two responses that set its outer dates are the two that no longer qualify. Everything here was collected between 5 and 15 May 2026.

What this is, and what it is not

The instrument. The Profit Check is a free online self-assessment of 14 questions, two for each of seven dimensions: profitability visibility, pricing and margins, tools and governance, strategic decision support, data and technology, cost allocation, and process and capacity design. Each question offers four answers scored 0 to 3. A dimension score is the average of its two questions expressed out of 100, so a dimension can only take the values 0, 17, 33, 50, 67, 83 or 100. The overall score is the average of all fourteen, out of 100, and the bands the tool reports back are Basic 0 to 25, Developing 26 to 50, Integrated 51 to 75 and Optimized 76 to 100. It takes 12 to 15 minutes and asks for no email address.

How these organizations arrived. Nobody was invited, sampled or paid. Every one of them found the assessment on this website and chose to take it. That is the central limitation of this report and it cannot be corrected by collecting more of the same: people who measure their own costing are more likely to already suspect a problem with it, so these figures almost certainly sit below whatever the true population average is. Read every number here as a statement about 33 finance teams who took this assessment, never as a statement about companies in general.

What is counted. A response is in this report when all six of these hold: it was taken on the version of the assessment in use up to 3 June 2026; it did not come from an address at a domain this firm owns or from the founder's own addresses; it names an organization and carries no test marker; the assessment's own quality check did not flag it as straight-lined, bot-paced or junk; all seven dimensions were scored; and where one organization submitted more than once, only its first response counts. Applying that rule to everything the assessment has ever stored leaves 33 responses.

What is suppressed, and why. We publish no cell of fewer than five organizations. A sector cut with two or three respondents identifies them to anyone who works in that sector, and this sample is concentrated in one country, which makes small cells more identifying rather than less. The suppressed responses are still inside every total on this page. We publish no free-text answer, no organization name and nothing about any individual respondent, and we never will.

What this cannot support. It cannot support a sentence of the form "X per cent of companies". It is not a random sample, it is not weighted, it has no confidence interval, and at n = 33 a difference of two or three points between two dimensions means nothing. It can support a sentence of the form "X of the 33 finance teams who took this assessment", and that is the form every finding above is written in.

We publish the method so the numbers can be read for what they are. The next edition needs responses on the revised assessment, not more of these. When there are enough of them to clear the same threshold, we will publish them and say what moved.

Frequently asked

Go deeper